Follow Datanami:
December 10, 2012

Ex-Cons Convert for Big Data

Ian Armas Foster

Convex Computing Corporation was a major supercomputing vendor in the 80’s and early 90’s before getting bought out by HP in 1995. The company was founded in 1982 and took on the supercomputing giants of the time (who incidentally remain notable today) in Cray, IBM, and SGI. They really placed itself on the map when it built the world’s first supercomputers that cost less than a million dollars.

A few of the former Convex higher-ups (they call themselves “Ex-Cons”) formed a new company in Convey, which is trying to take on four data-intensive computing sectors at once: life sciences, government, research and big data.

 “We’ve put the band back together,” said Convey CEO Bruce Toal. His “Ex-Con” complotters include Chief Scientist Steve Wallach, and Chief Technological Officer Tony M. Brewer. “We know each other’s strengths and weaknesses,” said Toal of the 52-man team, 60% of whom worked for Convex in its heyday.

The new company found movement in investors, with Convey having raised $40 million from firms like CenterPoint Ventures, Intel Capital and Xilinx in seed money before their last round of funding in 2009. With help of that funding, Convey is able to tackle all four sectors at once through chips called Field Programmable Gate Arrays (FGPAs).

According to Intersect360’s Addison Snell, these FPGAs enhance computing done on Convey’s x86 processors. “They have essentially tailored the system for applications where a substantial portion of the calculations are offloaded to the FPGAs,” said Snell “thus easing the burden of x86 processors.”

While they are officially focused on the afore-mentioned sectors, big data ties them all together, as government, research, and life science technological initiatives rely increasingly on big data to succeed.

“There are a lot of different dimensions to big data,” Snell said. “Sometimes the size of the file is huge. Sometimes the number of inputs is huge… If you’ve got this type of architecture that Convey has, there’s a huge opportunity to find more areas that are in or beyond, the areas of high-performance computing affected by big data.”

Specifically, Fox, Convey’s new computing architecture, reduces how much data has to travel between the processor and the coprocessor. As a result, Convey’s newest machine can reportedly run 4,000 threads in parallel compared to their previous amount of 1,000, while holding about four terabytes in shared system memory.

As the big data demand expands, so too will the competitive fire of vendors looking to plant their names near the top of the industry. Toal has proven that he can take on the big guns once, and according to Snell he has the drive to do it again. “In enterprise computing, if it works, you don’t touch it. In HPC, if it works, it’s no longer interesting. Presumably Bruce has a ‘Convez’ in him somewhere along the way,” said Snell.

Related Articles

20 Lessons Enterprise CIOs Can Learn from Supercomputing

IEEE Tags Top Tech Trends of 2013

Nine Databases in 45 Minutes

 

Datanami